Goblin Squad Member
|
Minor update: Roles should now require minimum ability scores. Also fixed a minor display issue for suggestions when you have already trained everything.
Edit: Is bugged for rogues. The fix is pretty simple so I will be releasing v18 later today with other updates and a fix to this.
|
Goblin Squad Member
|
v18 changelog:
==========
*Fixed bug with rogue ability scores.
*Roles should now be included in prerequisites (both in display windows and for training purposes). Outside of 'or' options to be added in roles the coding for roles is now complete.
*Some 'inf' suggestions giving no increase now listed as 0 at bottom of suggestion list instead.
*Fixed bug with 'OR' prerequisites in feats such as Elven Weapon Proficiency.
*'OR' prerequisites should now be showing with an 'or' in message windows for feats and roles.
*Message windows can be saved to text files. These files will appear in the saved_files folder. Currently there is no way to change the name of the output file name. The filename is based on the title of the window.
Note that there are circular prerequisites for high level fighters making Fighter 19+ unachievable.
|
Goblin Squad Member
|
| 1 person marked this as a favorite. |
v19 changelog:
===========
*Can now import and export training plans. These plans are simply lists of feats in the order they are to be trained and are saved in the training_plans folder. This should be useful for mature builds which have already trained many feats and wish to train more without having to train up from scratch each time. Players can then share these plans with eachother.
*Importing the plan will reset the current training and attempt to train everything in the plan.
*Sample plans to train up to wizard 1 and cleric 1 are included.
*Exported plans will include everything trained so far in order.
|
Goblin Squad Member
|
v20 changelog:
===========
*Reset on the training tab now also resets the input xp cost to 0.
*Saving message windows to text files now include commas to separate entries.
*Added an early version of a wiki tab to look up the details of attacks. This should help in figuring out which attacks correspond to which weapons when training. I don't know if this will be helpful for others, but I find it handy to know which attacks correspond to certain weapons when training.
Unfortunately, the information in some of the attacks will be out of date as I included an older file in this build.
|
Goblin Squad Member
|
Note that Dungeoneering, Nature, and Religion skills are being changed from Int to Wis, and History and Local are changing to Per. It's easy enough to edit the CSVs in LibreOffice for anyone who wants to make that change.
Because of non hard-coded input any change made in the input files should propagate to the calculator the next time you open it. This should allow others to update the input files for me in case I forget to.
|
Goblin Squad Member
|
Regarding input file cleanup, I made the following changes to the input files that you may wish to consider for future versions:
- Attacks.csv: Removed Basic Attacks and all entries that started with "Club" or "Hammer".
- Feature.csv: Removed Double-Weapon Specialization, Firearms Specialization, and Flail Specialization.
- Proficiencies.csv: Removed Double-Weapon Proficiency, Firearms Proficiency, and Flail Proficiency.
- Skills.csv: Removed Architect, Carpenter, Stonemason, SenseMotive, HandleAnimal, Athletics, DisableDevice, EscapeArtist, Fly, Perform, Persuasion, and Ride.
- Skills.csv: Changed Dungeoneering, Nature, and Religion to Wisdom and History and Local to Personality.
|
Goblin Squad Member
|
Right now I've only been taking the input files directly from raw dev files. If someone wants to post edited input files here please do! (This is something I've been hoping for long term.)
As far as efficiently training up to a target feat, that may be tricky to code up for the general case. I'd have to think about it for a while before knowing if it's viable to include. I suspect it's probably possible, but a bit of work.
Eg. I need 16 int. I have 15.99 int. Sorting feats by efficiency in training in int (most increase/xp) is the wrong way to go in that case. In that case I'd need to check which one cost the least xp, not the least xp per increase. So in practice finding the most efficient way to train something may involve many possibilities to check. Edit: Also, later there will likely be many feats which require levels in one role or another. These also have to be checked.
In terms of efficiency of ability score/xp spent, that's exactly what the "suggestions" button is for. Select an ability score from the optionmenu beside the button and then press the suggestions button to see a pop-up window of feats which increase that score. That pop-up is sorted in terms of efficiency.
|
Goblin Squad Member
|
I was thinking more of a scenario where I want to get Heavy Armor Proficiency 2 for the least amount of XP. It would be nice to select that feat and level, press a button, and have a build created that obtains the selected feat at the selected level for the least amount of XP.
That is essentially what I'm doing now by bringing up the Suggestions window and manually training every item on the list, in a row, one at a time.
|
Goblin Squad Member
|
| 1 person marked this as a favorite. |
That's what I was responding to above. In simple cases it's easy, but can quickly become complicated. Easy to describe, trickier to code.
Eg. if fighter 8 or paladin 8 were required to train Heavy Armor Proficiency then you need to check both of those roles. Then what if one of the prereqs for fighter 6 requires fighter 4 or barbarian 4? Depending on how complex the branching becomes as new roles are added this can get nasty. What about some feats that require Fighter 19 or Barbarian 19? It gets even worse. Right now in alpha it's easier because of a lack of overlap in roles, but it won't always be that way.
Eg. I need to raise multiple ability scores. Does the algorithm raise them one at a time? What about feats which raise multiple scores?
Eg. my above example where the most efficient (efficiency = increase/xp) feat isn't the ideal choice as you can reach the ability score goal with another feat which is less efficient, but costs less total xp. (ie. the more efficient feat overshoots the goal).
Then there's the question of how do you know that way is the least xp when there's multiple possible paths? Well, you have to compare the costs of each path. If the algorithm is slightly inefficient that could potentially wind up with many time consuming possibilities which wouldn't run so well on low end machines.
A rough idea of an algorithm is:
1) make sure the feat exists
2) make sure that the requested level isn't above max level that feat can be trained to. If so, set the requested level to that max level.
3) make a list of prerequisite feats and the prerequisites of those prerequisites on down the chain. Edit: Some feats have 'OR' prerequisites, so even here there's potentially multiple lists. Some of those 'OR' options include racial specific feats (eg. Elven Weapon Proficiency), so the algorithm may require a race as input as well.
4) at same time make a list of prerequisite roles
5) (for each role) add list of prerequisite feats for that role to the list of prerequisite feats. Potentially this can lead to many different lists at this point already depending on which prerequisites require multiple possible roles.
6) (for each list) compare to trained feats and remove any feat already trained to sufficient level from that list
7) attempt to train up each required feat, removing from list as completed. This may require a few cycles through as some of them may be prereqs for eachother. Break the cycle when nothing is trained in a given loop.
8) untrained feats in the list must be short of ability score, so make a list of those short abilities, sorting in order of which is the most short of its goal
9) somehow deal with the fact that the most efficient feats are not always the best choice (see example above with overshooting) and train scores up. What about feats which can raise multiple scores? This leads to multiple different possibilities here for every one of the above possibilities.
10) after each score trained is up to the next integer, try training each feat remaining (go back to the loop in 7).
Then hope that the program exits the algorithm cleanly and doesn't accidentally get stuck in an infinite loop. Step 9 is the big worry, along with the possibility of training from multiple roles and feats which raise multiple ability scores.
|
Goblin Squad Member
|
I have modified all the feats files to remove anything that is not currently trainable or is not expected to be implemented in this next patch (Alpha 8). For the modified versions, CLICK HERE.
The modifications include the following:
- Attacks.csv: Removed Basic Attacks and all entries that started with "Club" or "Hammer".
- Bonuses.csv: Removed Encumbrance Bonus
- Feature.csv: Removed Crossbow Specialization, Double-Weapon Specialization, Firearms Specialization, and Flail Specialization.
- Proficiencies.csv: Removed Crossbow Proficiency, Double-Weapon Proficiency, Firearms Proficiency, and Flail Proficiency.
- Skills.csv: Removed Architect, Carpenter, Stonemason, SenseMotive, HandleAnimal, Athletics, DisableDevice, EscapeArtist, Fly, Perform, Persuasion, and Ride.
- Skills.csv: Changed Dungeoneering, Nature, and Religion to Wisdom and History and Local to Personality.
|
Goblin Squad Member
|
For those who want to use the above, simply place them in the feat_files folder and they'll be read in next time you open the calculator.
For those wanting to edit the input files, there is a simple way that you can save time with changing the ability scores related to feats. The way feats look for ability score minimums/increases is based on some older coding than the way that prereqs are looked for. So for the time being if you change the ability scores needed for level 1 it should (untested! edit: did a simple test and appears to work) be interpretted by the calculator as being changed for all levels. This is because I assumed at the time that a feat wouldn't increase strength one level and then wisdom the next. I want to eventually change this to the same way that prerequisites are done, but it's a low priority.
|
Goblin Squad Member
|
I think that finding the way to maximize a skill with a given XP budget might be a NP-class problem. Fortunately we have a relatively small O right now.
I think I agree, (but only because NP contains P, i'm not fully convinced about the N bit).
It should be "fairly straighforward" to calculate the most xp-efficient route to a given ability score treshhold. The "but i just need 0.01 more" complication only applies to the final stage, so that's just a one-time add-on.. or wait... duh! I see your point.
(gah, couldn't we have had this discussion 10 years ago, when I was into this type of stuff)
ps: but is it the right problem? isn't more relevant to minimize costs for a given skill level? Since costs scale the way they do it's very hard to 'squeeze in an extra skill level at the top' through efficient budgeting.
|
Goblin Squad Member
|
Just waiting for the new spreadsheets before releasing v22. In the meantime here's a preliminary changelog for v22:
*Improved the sorting of passive feat names on wiki tab by making single digit levels 2 digits, eg. "Archer 3" -> "Archer 03". This is a temporary solution until I merge passives of the same name but different level.
*Added in specific achievements, eg. "Arcane Expert". These may have been in previously, but were taken out as they are mixed in with roles in the spreadsheets. With the roles there were potentially cases where some roles as feat prerequisites may have been missed due to the way the parser worked in the last few versions. Even though I don't know of any specific occurence of it, that potential bug should now be taken care of.
*Added a column to suggestions message window to show the xp cost for next level. Still sorted by increase per xp though.
|
Goblin Squad Member
|
One thing I've noticed in the Suggestions window is that repetitions of a given feat at a higher level are not included as a suggestion when I think they should be. So if Hit Points 3, for example, is the least XP-per-increase item to raise Con, then Hit Points will be at the top of the list but if Hit Points 4 is a better buy than the tenth item in the list, Hit Points will NOT be repeated on the list in that tenth spot because Hit Points has already appeared on the list.
The only way this can be addressed in the current build is to close the Suggestions window and relaunch it every time I train a feat that could potentially be repeated at a higher level, which is virtually everything.
|
Goblin Squad Member
|
| 1 person marked this as a favorite. |
Just waiting for the new spreadsheets before releasing v22.
Pathfinder Online (Public) -> "Raw from the Devs" -> "PFO Wiki - Official Data"
That's the official source of record, publicly shared by Stephen Cheney. He has already selected the "Publish to the web" option, so it's available via ListQuery and ListFeed objects in the Google Data API.
I've probably got another 4 or 5 hours before I can get the formatted stuff posted.
|
Goblin Squad Member
|
Ya, that's why it says next level only and as is needs to be reopened repeatedly. I suppose it's possible to add in all higher levels to the suggestions, but then it gets really long (eg. there are 40 levels of Power, 20 levels each of several skills, 6 levels each for dozens of attacks etc.) so I'd rather not add all levels in.
Long term I'd like to be able to refresh the info on there as you level up various feats, but haven't coded that up yet. Too early to say what form that refresh would take.
|
Goblin Squad Member
|
Not sure if this has been mentioned, but the ability to undo a feat training would be very useful. Right now, you can export the training.txt and erase the feats you don't want, save it, then re-import it, but that is wearisome.
Having said that, the newest build is FANTASTIC! I see myself using this program a lot in the coming years.
Thank you for all your hard work, Nightdrifter!
Daz
|
Goblin Squad Member
|
New v22 changelog:
===============
*Improved the sorting of passive feat names on wiki tab by making single digit levels 2 digits, eg. 3->03. This is a temporary solution until I merge passives of the same name but different level.
*Added in specific achievements, eg. "Arcane Expert". These may have been in previously, but were taken out as they are mixed in with roles in the spreadsheets. With the roles there were potentially cases where some roles as feat prerequisites may have been missed due to the way the parser works. That potential bug should now be taken care of.
*Added a column to suggestions message window to show the xp cost for next level. Still sorted by increase per xp though.
*Fixed a bug where importing training plans was causing the output training plan to be doubled up.
*Added a button to undo the last trained feat. This can be used repeatedly to undo as much as you'd like. In practice this simply retrains you from scratch up to (but not including) the last feat you trained, so for very xp heavy builds it may take a little time.
Goblinworks Executive Founder
|
I think that finding the way to maximize a skill with a given XP budget might be a NP-class problem. Fortunately we have a relatively small O right now.I think I agree, (but only because NP contains P, i'm not fully convinced about the N bit).
It should be "fairly straighforward" to calculate the most xp-efficient route to a given ability score treshhold. The "but i just need 0.01 more" complication only applies to the final stage, so that's just a one-time add-on.. or wait... duh! I see your point.
(gah, couldn't we have had this discussion 10 years ago, when I was into this type of stuff)
ps: but is it the right problem? isn't more relevant to minimize costs for a given skill level? Since costs scale the way they do it's very hard to 'squeeze in an extra skill level at the top' through efficient budgeting.
Yes, but the build that maximizes skill for N XP might be wildly divergent from the build that maximizes skill for N+Y XP.
And like lots of NP problems, many people can intuit very good solutions, but it is very difficult to prove that a given solution is optimal.
|
Goblin Squad Member
|
Updated changelog (apologies for spamming this as I keep adding changes to this version):
*Improved the sorting of passive feat names on wiki tab by making single digit levels 2 digits, eg. 3->03. This is a temporary solution until I merge passives of the same name but different level.
*Added in specific achievements, eg. "Arcane Expert". These may have been in previously, but were taken out as they are mixed in with roles in the spreadsheets. With the roles there were potentially cases where some roles as feat prerequisites may have been missed due to the way the parser works. That potential bug should now be taken care of.
*Added a column to suggestions message window to show the xp cost for next level. Still sorted by increase per xp though.
*Fixed a bug where importing training plans was causing the output training plan to be doubled up.
*Added a button to undo the last trained feat. This can be used repeatedly to undo as much as you'd like. In practice this simply retrains you from scratch up to (but not including) the last feat you trained, so for very xp heavy builds it may take a little time.
*Multiple behind the scenes changes to feat_reader to deal with some slightly different input for feats (essentially cases that hadn't come up before that the previous reader couldn't handle).
*Orisons spreadsheet was missing an AbilityReq Lvl2 column so I added one. Was causing issues with feat_reader.
*Added column titles into Armor spreadsheet (the feat_reader looks for these).
*Expendables training and roles are not updated.
*Consumables added to the wiki tab with expendables.
|
Goblin Squad Member
|
Role files (ie. what's needed to get levels in roles) and expendables training (xp and prerequisites related to expendables) have not been updated in this version. From early looks at the new role files the role_reader will need an overhaul to read them properly.
Minor progress in updating the suggestions windows: I can check if those windows exist, but haven't added in any needed functionality to make updates.
Edit: For now I have a private build set so that training a feat will automatically close and reopen any open Suggestions windows with the new information. You'll have to resize the windows every time, so I'm not sure if this is annoying or handy. Thoughts?
|
Goblin Squad Member
|
Okay, will keep that in for now.
Currently adding a quasi-search function to the wiki tab for expendables (attacks to be added afterwards). The idea being to have the user define a selection of keywords and the output will be a message window which displays a list of expendables sorted by the number of matched keywords. That way you can know which ones best match the keywords on your class feature or weapon. Don't yet have a way to discriminate between major and minors though, though that will only matter for attacks.
|
Goblin Squad Member
|
| 2 people marked this as a favorite. |
In the interests of getting to bed at a reasonable time for once I'm going to release the current version with only searchable expendables.
v23 changelog:
===========
*Any open Suggestions window will be closed and reopened with refreshed information upon training any feat. Unfortunately this will require resizing by hand.
*On the wiki tab you can select from expendable keywords and add them to a list. Pressing 'Find Expendables with Keywords' will bring up a message window with a list of expendables sorted by the number of matching keywords. (Attacks to gain similar functionality as I get time.)
|
Goblin Squad Member
|
Just to give people an idea of where this is going and to get some input, here is a list of upcoming features I want to add (how long it will take to add any given feature is highly variable so I emphasize the want):
*change wiki search to use class features to search expendables based on compatible keywords
*add weapons to wiki tab and search attacks based on their keywords
*add time to train on feat info window
*merge passives of different levels into a single passive on wiki tab that displays all level info
*behind the scenes achievement logic improvements
*behind the scenes role logic (to deal with the new role files that I haven't added and to deal with the eventual 'OR' logic for roles). (highest priority, but may not be done first)
*merge minor tabs (eg. move xp tab onto training tab). No functionality lost, simply moved.
*change defaults in reputation tab to reflect the new reputation scaling (needs dev input).
*reset button on threading tab
*a proper readme file, not a todo list disguised as a readme :p
*detailed instructions on how to update inputs
*simple stealth calculator
Longer term:
*crafting tab (features as yet undetermined)
*combat tabs (features not finalized, but should expand upon my older combat work)
*settlement DI tab (features as yet undermined)
Are there other changes anyone might suggest?
|
Goblin Squad Member
|
v24 changelog:
===========
*Updated expendables training. (New file was not available when v23 was released.)
*Tweaked the placement of buttons for searching expendables keywords to make your list more visible.
*Behind the scenes changes to roles to handle slightly modified format. Note that wizards have an 'or' in their ability scores.
*New roles files.
Training roles with the new logic has not been extensively tested, so please let me know if you find bugs.
|
Goblin Squad Member
|
Busy day...
v25 changelog:
===========
*Minor tweak in behind the scenes role logic to deal with potential corner cases.
*Trained feats will now show the xp cost for the next level of a feat.
*Added various new training plans as examples. (Users can make their own and share them. They can be found in the training_plans folder.)
*Added time to train each level of a feat and cumulative training time to feat info windows. Information is displayed as days/hours/minutes/seconds. If you change the xp rate in the Advanced tab (using "Store XP Settings" ) these times will be recalculated for that new xp rate when you open the feat info window again.
|
Goblin Squad Member
|
One more update for today:
v26 changelog:
===========
*Fixed bug where changing xp rate and then setting back to default would not propagate the change back to default to feat training windows.
*Merged passives of different levels into a single passive on the wiki tab which displays all level information.
*Changed wiki search for expendables based on keywords to use class features and a level as input. This should be more appropriate as expendable keywords are matched to class features.
*Wiki search for expendables based on keywords now also displays implement type used by the expendable.
|
Goblin Squad Member
|
v27 changelog:
===========
*Allowed the ignoring of feats to make editing csv files easier for cases of feats not trainable in game. Any feat listed in ignored_feats.txt will not be read in and will not show up in the calculator. (I may not have added the most complete list, so can someone please post a more comprehensive list of what goes in here?)
*Searches for expendables by keywords now also shows the level of the expendable.
|
Goblin Squad Member
|
v28 changelog:
===========
*Minor changes to saved text from message windows to (hopefully) make it easier to tell which entries belong to which column.
*Message windows now have check boxes at the bottom of each column. If you uncheck a column that column will no longer be included when saving as a text file. This allows the user to make text files without columns they don't want.
*Can now search based on expendables in the wiki tab. The search will look for appropriate class features to use with those expendables. If at least one of the expendables has a keyword matching the keywords in the class feature it will be displayed in a message window. The information displayed is the number of matching keywords.
|
Goblin Squad Member
|
V29 CHANGELOG:
===========
*Sorting of rows in searches for appropriate class features is now based on the sum of matches along a row.
*Expendables used in searches should now be sorted by alphabetical order.
*Can now save message windows as csv files (delimiter ; quotechar blank) which should be readable by spreadsheet programs.
*Minor tweaks in search by expendables to protect against corner cases.
|
Goblin Squad Member
|
Was poking around on my computer and dug out some old programs and added them to the calculator. Have a few more hanging around somewhere.
V30 CHANGELOG:
===========
*Added a new stat conversion tab to convert between some basic combat stats. Note: this is still in an early state and does not take into account secondary effects. May give weird results in corner cases (such as base damage less than resistance).
*This new tab also allows the calculation of roughly how many hits it would take to bring down the target. Again, no secondary effects.
